The Research Question
- Descriptive data
- Explain empirically why this is so à women seem more religious than men
- Different roles and different position occupied by man and women in society à structural location theory
- Narrowed research focus down to: which structural location factors accounted for greater religiousness of women à child rearing role / workforce participation / family focus.
Hypotheses
Workforce participation
Gender Child Rearing Stage Religiousness
Family Focus
- In order to test each of the 3 paths – at least 7 hypotheses can be developed:
o Differences between gender in religious orientation will disappear once the ffect of child rearing cycle is removed.
o Among men and women who have no children and those who children are grown up there should be fewer religious differences than among men and women still in active child rearing stage
o Women in the workforce will be less religious than women out of the workforce
o The religious orientation of women in the workforce will be more similar to males in the workforce than to females out of the workforce
o When effects of gender differences in workforce participation are removed / controlled, gender differences in religious orientation will be reduced markedly.
o Those who are focused more on the family will be more religious than those with less family focus
o When effects of gender differences in family focus are removed, gender differences in religious orientation will be markedly reduced.
Clarifying and Operationalising The Concepts
- Secondary data: 1983 Australian Values Study Survey
- N = 1228
- Variables
o DV – religion
o Independent and intervening variables
§ Gender
§ Workforce participation
§ Stages in child rearing
§ Family focus
o Background variables
§ Age
§ Country of birth
§ Family income
§ Occupation
- Religion
o Correlation matrix pp187-8
- Independent & intervening variables
o Developed composite variable
o Collapsing categories of existing variables
Preparing variables for Analysis
- Reduce number of variables to manageable number
- Proper uni dimensional scales required to reduce number of variables to be analysed
- Ensure items in each scale belonged together
- Each variable in each scale scored in same direction
- Had to construct scales so that scoring or number of categories of one variable did not bias overall scale
- Missing data à used mean
- Prepare variables for regression analysis
Initial Analysis
- Convention à scales based on summated ordinal questions can safely be treated with interval level statistics
o Kendall's tau & Spearman's rho
o Cross tabulations with gamma
o Comparing means
Testing the Explanations
- Difficulty with cross tabulation analysis à have not controlled for other variables
o More variables that are controlled à more we confront problems with small sample sizes and stat reliability
- Multiple regression analysis
Discussion
- Workforce participation is important explanatory variable, but not clear precisely why this is so
-
No comments:
Post a Comment